In biology classrooms, we often learn that genes are transcribed into mRNA, which is then translated into proteins—a tidy, linear process. But gene expression in living cells is far more intricate, governed by a complex network of molecular players. Among these, transcription factors (TFs) have long been considered the master regulators—the proteins that “read” the DNA blueprint and influence which genes to turn on or off.
But what if those “architects” aren’t actually following the blueprint as closely as we thought? That’s exactly the conclusion of a new study in Nature from the Hahn lab led by postdoc Dr. Lakshmi Mahendrawada and Dr. Steven Hahn. By mapping where transcription factors bind in yeast—and how that relates to gene activity—they uncovered a twist: where TFs bind, and whether they recognize their usual DNA sequence motifs, often doesn’t line up with the genes they regulate.
Transcription factors constitute a large family of proteins—about 150 in yeast and over 1,600 in humans—that regulate gene expression by binding specific DNA sequences and, often, recruiting co-factors to initiate transcription. To map where these proteins interact with DNA throughout the genome, the researchers used an advanced technique called chromatin endogenous cleavage sequencing (ChEC-seq).
ChEC-seq is a powerful method—developed by the Henikoff lab here at Fred Hutch—that tags each TF with micrococcal nuclease, an enzyme that cuts DNA only where the TF binds. This precise DNA cleavage, triggered by calcium, enables researchers to pinpoint TF binding sites across the genome with minimal background noise, offering an important improvement over older techniques.
Surprisingly, the study found that the presence of a TF’s canonical DNA motif—the short sequence it supposedly “recognizes”—had little predictive power for actual binding. Instead, these results and other indicate that factors like chromatin accessibility, interactions with cofactors, and local DNA shape may play a bigger role in determining whether a TF bound at a site. In other words, the “right sequence” alone isn’t sufficient; the local DNA context and chromatin environment matter greatly.
The researchers further categorized yeast genes based on how many transcription factors bound near each promoter, revealing distinct clusters. Many promoters had multiple TFs bound at once—sometimes dozens—suggesting that transcription factors often act in concert at the same regulatory regions. This cooperative binding hints that gene regulation operates more like a coordinated network than a series of isolated, one-on-one interactions.
To probe the underlying mechanisms, the researchers focused on Gcn4, a well-known TF that responds to amino acid starvation. By mutating either its DNA-binding domain or its activation domains—which normally recruit other proteins to initiate transcription—they found something unexpected: Gcn4 still associated with many of its usual genomic sites even when its DNA-binding domain was disabled. This suggests that the activation domain can help stabilize Gcn4’s binding, likely by recruiting cofactors that alter the local chromatin environment to make DNA more accessible. This flips the traditional model, where TFs were thought to bind DNA first and then recruit cofactors. Instead, activation domains can facilitate binding in the first place, revealing a more dynamic interplay between TF structure, cofactors, and chromatin state.
This cooperative binding hints that gene regulation isn’t just a straightforward reading of a DNA blueprint by lone architects. Instead, it’s more like a bustling construction site where multiple architects and workers coordinate dynamically, sometimes improvising based on the local environment and who's on site. This means the “blueprint” isn’t always followed literally — context and teamwork shape the final design.
While ChEC-seq reveals where TFs bind, it doesn’t tell us whether binding actually changes gene expression. To connect binding with function, the team employed an auxin-inducible degron system to rapidly degrade 126 different TFs individually, and then measured the immediate effects on newly synthesized RNAs via RNA sequencing.
But before jumping into the results, it helps to understand that yeast genes fall into two major regulatory classes. Some are transcription factor IID-dependent genes, which rely mainly on the TFIID complex to recruit RNA polymerase II and kickstart transcription. Others are called coactivator-redundant genes — these are more flexible and can be activated through multiple pathways. This classification provides a framework for teasing apart how different genes respond to losing specific transcription factors.
Upon degrading these individual TFs, expression of nearly 5,000 genes changed. After filtering out indirect or stress-related effects, about 4,725 genes were identified as directly regulated by one or more TFs. Interestingly, many genes were influenced by multiple TFs, with some affected by as many as 21 regulators. Coactivator-redundant genes tended to be controlled by a greater number of TFs and highly regulated.
Most TFs had focused effects, regulating fewer than 200 genes. Others fine-tuned gene expression in specific biological contexts. But a few coordinated widespread transcriptional programs, including master regulators Cyc8, Rap1, and Abf1 that influenced over a thousand genes each. Most surprising was that in many cases, genes responded to TF depletion, but TF binding was undetectable near the regulated gene.
This study reframes gene regulation as a dynamic, context-dependent process. Binding is not dictated solely by DNA sequence motifs but involves chromatin accessibility, TF cooperation, and domain-specific functions within TFs themselves. Furthermore, the functional impact of binding—whether a gene is turned on or off—depends on complex regulatory networks rather than simple one-to-one relationships. “Surprisingly, TFs rarely function strictly as activators or repressors. Instead, most have dual roles,” shared Mahendrawada.
So, where do we go from here? According to Mahendrawada, plenty of big questions remain. How do transcription factors regulate genes without binding nearby? Many TFs also play a dual role, sometimes turning genes on and other times turning them off — but what controls this switch? Is it structural changes of the TF, post-translational modifications, or the company they keep on the DNA?
Plus, how do factors like chromatin state and promoter architecture influence whether a TF activates or represses a gene? With so many TFs working together, could machine learning—which the Hahn lab has already used to help understand transcription—help us predict gene regulation by analyzing binding patterns and DNA features?
Exploring these questions will deepen our understanding of gene regulation’s complexity—and undoubtedly reveal even more surprises along the way.
Fred Hutch/University of Washington/Seattle Children’s Cancer Consortium Member Dr. Steven Hahn contributed to this research.
The spotlighted research was funded by National Institutes of Health.
Mahendrawada L, Warfield L, Donczew R, Hahn S. 2025. Low overlap of transcription factor DNA binding and regulatory targets. Nature. https://doi.org/10.1038/s41586-025-08916-0.